Thumbsnag crawls an HTML document and finds imagery that best represents the given page.

use Thumbsnag\FastImageAnalyzer;
use Thumbsnag\Thumbsnag;
use Thumbsnag\UrlDocument;

$url = 'http://simplegifts.co';
$html = file_get_contents($url);

$document = new DOMDocument();
$document->loadHTML($html);

$analyzer = new FastImageAnalyzer(new FastImage());

$thumbsnag = Thumbsnag::load(new UrlDocument($doc, $url), $analyzer);
$images = $thumbsnag->process();

After inspection, $images will return something like:

Array
(
  [1] => Thumbsnag\Image Object
  (
    [url:Thumbsnag\Image:private] => http://simplegifts.co/image1.jpg
    [height:Thumbsnag\Image:private] => 565
    [width:Thumbsnag\Image:private] => 849
  )

  [2] => Thumbsnag\Image Object
  (
    [url:Thumbsnag\Image:private] => http://simplegifts.co/image2.png
    [height:Thumbsnag\Image:private] => 450
    [width:Thumbsnag\Image:private] => 1162
  )
)

Pull the package in through Composer.

"require": {
  "gstjohn/thumbsnag": "~1.0"
}

Thumbsnag::load() takes an array as its third parameter which overrides the default config. Available configuration options are:

• min_area (default: 5000)

  This option represents the minimum pixel area (width x height) required in order to be included in the result set.

• ratio_threshold (default: 3.0)

  This option represents the maximum ratio of width-to-height allowed in order to be included in the result set.

• filename_filters (default: "sprite", "blank", and "spacer")

  This option represents an array of words that must not be in the image file name in order to be included in the result set.

• How to Find the Image That Best Represents a Web Page
• Crawlsane

Thumbsnag is open-sourced software licensed under the MIT license.